Anthropic has made the unprecedented decision to globally disable access to its most advanced Claude models—Fable 5 and Mythos—following an explicit order from the US government targeting foreign national users. The directive, framed around national security concerns, represents a dramatic reversal of the company's previous approach to international access and reflects growing US government scrutiny over advanced AI capabilities. This action underscores escalating tensions between Anthropic and federal authorities who have expressed concerns about supply chain vulnerabilities, potential jailbreaking risks, and the unauthorized proliferation of cutting-edge AI technology beyond US borders.
The US government's intervention centered on multiple risk categories beyond simple foreign access restrictions. Officials cited concerns about supply chain compromises and the possibility that adversaries could exploit these models to develop jailbreaking techniques that might circumvent safety guardrails. The restrictions specifically target Anthropic's top-tier offerings, suggesting officials view these models as particularly sensitive national assets that require controlled access. By mandating this global ban rather than allowing Anthropic discretionary implementation, the government signaled the severity with which it views the security implications.
This development carries substantial implications for Anthropic's international operations and the broader AI industry's relationship with government oversight. The company faces a challenging balance between complying with US security directives and maintaining its position as a trusted global technology provider. The move may establish a precedent for how advanced AI developers manage access to frontier models, potentially reshaping international AI governance frameworks. For Anthropic, the incident demonstrates that even leading safety-conscious AI companies face regulatory pressures that can fundamentally alter their business models and product accessibility worldwide.